BASIC RESEARCH www.jasn.org Quantitative Proteomics of All 14 Renal Tubule Segments in Rat Kavee Limbutara, Chung-Lin Chou, and Mark A. Knepper Epithelial Systems Biology Laboratory, Systems Biology Center, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Maryland ABSTRACT Background Previous research has used RNA sequencing in microdissected kidney tubules or single cells isolated from the kidney to profile gene expression in each type of kidney tubule epithelial cell. However, because proteins, not mRNA molecules, mediate most cellular functions, it is desirable to know the iden- tity and amounts of each protein species to understand function. Recent improvements in the sensitivity of mass spectrometers offered us the ability to quantify the proteins expressed in each of 14 different renal tubule segments from rat. Methods We manually dissected kidney tubules from rat kidneys and subjected samples to protein mass spectrometry. We used the “proteomic ruler” technique to estimate the number of molecules of each protein per cell. Results Over the 44 samples analyzed, the average number of quantified proteins per segment was 4234, accounting for at least 99% of protein molecules in each cell. We have made the data publicly available online at the Kidney Tubule Expression Atlas website (https://esbl.nhlbi.nih.gov/KTEA/). Protein abun- dance along the renal tubule for many commonly studied water and solute transport proteins and meta- bolic enzymes matched expectations from prior localization studies, demonstrating the overall reliability of the data. The site features a “correlated protein” function, which we used to identify cell type–specific transcription factors expressed along the renal tubule. Conclusions We identified and quantified proteins expressed in each of the 14 segments of rat kidney tubules and used the proteomic data that we obtained to create an online information resource, the Kidney Tubule Expression Atlas. This resource will allow users throughout the world to browse segment-specific protein expression data and download them for their own research. JASN 31: 1255–1266, 2020. doi: https://doi.org/10.1681/ASN.2020010071 The introduction of RNA sequencing (RNA-Seq) has can be independently regulated by processes that con- provided an infusion of new information about gene trol protein stability and translation.12,13 expression in the kidney. The method is extremely Consequently, there is a strong need for quanti- sensitive, allowing transcriptomic profiling of indi- tative proteomic methods with sensitivity that vidual renal tubules1 and single cells isolated from kidney.2–5 RNA-Seq data are very valuable in helping researchers identify hypotheses for further study, sup- Received January 17, 2020. Accepted March 9, 2020. plying what amounts to “instant preliminary data” Published online ahead of print. Publication date available at when provided as online resources. Yet, there is a ma- www.jasn.org. jor limitation in the sense that proteins, not mRNA Correspondence: Dr. Mark A. Knepper, National Heart, Lung, molecules, are responsible for most biologic func- and Blood Institute, National Institutes of Health, Building 10, tions in the cell. Several studies have demonstrated Room 6N307, 10 Center Drive, MSC-1603, Bethesda, MD 20892- that protein abundances are often not predictable 1603. Email: [email protected] – from mRNA levels6 11 because protein abundances Copyright © 2020 by the American Society of Nephrology JASN 31: 1255–1266, 2020 ISSN : 1046-6673/3106-1255 1255 BASIC RESEARCH www.jasn.org rivals that of transcriptomics. The problem has been that, Significance Statement although RNA-Seq benefits from PCR amplification, similar amplification is not possible for proteins. Instead, increased The renal tubule’s 14 distinct segments consist of epithelial cells sensitivity for mass spectrometry–based proteomics de- with different transport and metabolic functions. Identifying the pends on improvements in mass spectrometer sensitivity. proteins mediating each function is crucial to gaining an overall understanding of kidney physiology and pathophysiology. New 14 Largely because of such progress, Rinschen and colleagues developments in protein mass spectrometry have resulted in a have recently shown that it is possible to obtain deep pro- marked increase in sensitivity of protein detection and quantifica- teomes from microdissected renal tubules. Here, we use tion. In this study, the authors manually microdissected kidney tu- similar techniques to identify proteomes of 14 distinct renal bules from rat kidneys and leveraged the advances in protein mass tubule segments, working with microdissected tubules spectrometry to identify and quantify the proteins expressed in each of the 14 tubule segments. They used these data to create an from rats. online information resource, the Kidney Tubule Expression Atlas, to allow researchers throughout the world to browse segment- specific protein expression data and download them for their METHODS own investigations. Microdissection for 30 minutes. The modified single-pot, solid phase–enhanced We followed the previously published standard protocol for sample preparation protocol17,18 was used to clean and digest 1,15 rat kidney tubule segment microdissection. Concisely, proteins into peptides for mass spectrometry analysis. The re- – – male Sprague Dawley rats age 4 8 weeks (Animal Study Pro- sulting peptide mixtures were then fractionated using micro- tocol No. H-0110R4; approved by the Animal Care and Use scale basic reverse-phase liquid chromatography19 into eight Committee, National Heart, Lung, and Blood Institute) were fractions (elution buffers: 5%, 7.5%, 10%, 13%, 16%, 20%, euthanized by decapitation, and kidneys were perfused via 25%, and 30% acetonitrile in 10 mM triethylammonium bi- aorta with 10 ml of buffer solution (120 mM NaCl, 5 mM carbonate). Peptide mixture fractions were concatenated (frac- KCl, 2.5 mM Na HPO , 5 mM HEPES, 1.2 mM MgSO , 2 4 4 tion 1 with 5, 2 with 6, 3 with 7, and 4 with 8), resulting in four 2mMCaCl , 5 mM sodium acetate, 5.5 mM glucose, adjusted 2 fractions per sample (except one cortical collecting duct sample to pH 7.4 by NaOH, bubbled with 100% oxygen) to remove that was not concatenated). Fractionated samples were then the blood. The kidneys were then further perfused with 10 ml analyzed with LC-MS/MS using a Dionex UltiMate 3000 digestion solution containing the same buffer plus either nano HPLC system coupled with Orbitrap Fusion Lumos 1 mg/ml of collagenase B (Roche) for cortex or 3 mg/ml for (Thermo Scientific). Peptides were introduced into a nanotrap medulla. (Separate rats were used for cortical and medullary column at a flow rate of 300 nl/min and then separated on a dissections.) For medullary tissue, 1 mg/ml (outer medulla) or m 3 3 mg/ml (inner medulla) of hyaluronidase (Worthington Bio- reverse-phase EASY-Spray PepMap column (C18, 75 m 50 chemical Corporation) was also added to the digestion solu- cm) using a nonlinear gradient from 2% to 28% acetonitrile in tion. Kidneys were removed and cut into thin slices, and then, 0.1% formic acid (runtime 120 minutes). Data-dependent ac- they were incubated with the same digestion solution at 37°C quisition was performed with MS1 resolution of 120,000 and for 30 minutes (cortex), 40 minutes (outer medulla), or MS2 resolution of 15,000/30,000 at cycle time of 3 seconds. fi 90 minutes (inner medulla). After the digestion, kidney tubule All mass spectrum raw les were searched against a rat microdissection was performed under a Wild M8 stereomi- UniProt reference proteome (release 2019_10) using Max- 20 fi croscope. Each tubule segment was distinguished by its char- Quant 1.6.10.43. Parameters for amino acid modi cation fi acteristics as previously described.1 A short description of the including xed carbamidomethyl (C) and variable Oxidation recognition criteria is given as Supplemental Table 1. Tubule (M), Acetyl (Protein N-term). Match between run option fi length was measured.16 Several dissected tubules were pooled was enabled. Trypsin/P was con gured as digestion enzyme. together and transferred with 2 ml to a clean petri dish con- Default settings were used for other parameters. taining ice-cold PBS. Tubules were washed several times To estimate protein abundance in each sample, we applied with PBS and then lysed in 15 mlof1.5%SDS/100mM the proteomic ruler approach21 using an in-house Python triethylammonium bicarbonate/13 Halt protease and script. Briefly, protein intensities were summed for each sam- phosphatase inhibitor by pipetting up and down under a ple and used to normalize differences in total protein amount. stereomicroscope. Lysed samples were sonicated using a Molecular weight and number of theoretical peptides (tryptic cup horn probe (Misonix Sonicator 3000) for 5 minutes digested peptides with length 7–30 amino acids) were used to andkeptfrozenat280°C until further processed. correct for differences in signal intensity caused by protein size and sequence. On the basis of the assumption that total Mass Spectrometry–Based Proteomics amount of histones is approximately equal to amount of For each sample, several tubules were pooled together. Protein DNA,21 total normalized signal intensity of histones in each lysates were reduced with 10 mM dithiothreitol at 37°C for sample was used to estimate copy number per cell for every 30 minutes followed by alkylation using 10 mM iodoacetamide protein. Plotting the total histone signal against the estimated 1256 JASN JASN 31: 1255–1266, 2020 www.jasn.org BASIC RESEARCH number of cells from each sample gave a strong correlation in the IMCD. Expression levels were quantified as copies per (Supplemental Figure 1). The total cells per sample were esti- cell (cpc) using the Proteomic Ruler technique.21 Values were mated for segments other than thin descending limbs from successfully quantified over six orders of magnitude, a feat data for cells per unit length curated by Clark et al.22 and unachievable with immunochemical methods. The full data- multiplied by the total length of tubules in each sample.